Community Size-Structure of the Northwest Atlantic Groundfish Communities, Response to Direct Disturbance and a Changing Environment

Results from a Individual Size Distribution Analysis of the Northeast US Groundfish Community

Author
Affiliation
Adam A. Kemberling

Gulf of Maine Research Institute

Published

April 24, 2023

Potential Journals:

  1. ICES Journal of Marine Science

Abstract

The surface waters of the Northwest Atlantic Ocean are among the fastest warming on Earth. This area is highly-productive biologically, and there are concerns that ecological consequences will follow this rapid-warming. Research on the impacts of this rapid warming has primarily focused on high-profile and/or upper trophic level species. Ecological theory and laboratory studies suggest that elevated temperatures facilitate early maturation and smaller adult body-sizes. However, it is unclear whether that relationship might be mitigated against through adaptive behaviors in an open ocean environment. Here we’ve investigated ecosystem wide impacts on the individual size distribution (ISD) to track changes in community size structure. In cases where community responses are not adequate to counter the impacts of elevated temperatures, we anticipated a steepening of the size spectrum slope (ISD exponent). A steeper relationship relating to a reduction in larger sized individuals and an increased prevalence of smaller sized individuals. Using data from fisheries independent surveys we calculated the community size spectra for four regions along the US NE continental shelf. Correlation/regression analyses were then performed to then assess the degree to which these changes were in alignment to hypothesized bottom-up and top-down disturbances. At the regional scale, we found that community size structure changes (spectra slope) were the largest in the Northern regions, in the Gulf of Maine and Georges Bank. These areas are home to the coldest temperatures and the largest proportions of groundfish species in the community. Spectrum slope declines were most pronounced in the 80’s and 90’s, before the rapid warming of the last decade. The timing of these declines suggest that external factors drove the initial declines of larger-sized individuals within the communities, before elevated temperatures began to influence the ecosystem. Correlation analyses reveal that while fisheries landings are strongly correlated with these declines, bottom-up factors of zooplankton community metrics, Gulf Stream Index, and SST anomalies are also important. While the primary pressure of fisheries exploitation has declined dramatically over time, the recovery of larger-sized individuals has not been seen. That kind of recovery will likely depend on the elevated temperatures seen over the last decade.

Introduction

Normalized Size Spectra & Individual Size Distributions

Size is a defining characteristic of species and mediates many ecological interactions and metabolic pathways (Brown et al. (2000)) . Size is a big factor in determining the mobility of an organism. Mobility then impacts the ability to evade predation, foraging success, efforts to locate and follow essential habitats, geographic ranging, and the metabolic costs associated with each of these behaviors (Hillaert et al. (2018)). Body size also mediates vulnerability to aspects the immediate environment such as temperature through heat exchange or the threat of desiccation in terrestrial species (Gillooly et al. (2001); Heatwole et al. (1969)). Body size even informs life history features like life span and the trophic position an individual might occupy through its impact on metabolism and resource use (White et al. (2007)).

Size structured environments are a fundamental organizational pattern globally that emerges from these relationships add_citation. Within strongly size-structured ecosystems, growth and maturity changes alter fitness and ultimately determine whether a species is successful in that environment add_citation . Ecological theory is rich with models relating how energy transfers from smaller prey species to larger predatory trophic levels, the allocation of energy for growth, and the trade offs of allocating energy towards those ends (Bertalanffy (1938); Bertalanffy (1957); add_citations). A globally persistent pattern in ecology entangled in those relationships and their critiques is the decline in abundance with increasing body size (Damuth (1981); Currie (1993); Sheldon et al. (1972); Loeuille and Loreau (2006)).

This relationship, between size and abundance, integrates multiple processes operating on the cellular, individual, and community levels simultaneously. The quantities for size and abundance are also some of the most readily collected data assets of any ecological community. This creates an opportunity to learn much about a system from a relatively low-effort in data collection. For these reasons, size spectrum analyses and individual size distribution (ISD) methods have gained increasing attention as an entry point to assessing ecosystem health and to detect system-wide disturbance (Shin et al. (2005)‘; Pomeranz et al. (2022); White et al. (2007)). An advantage of these models is they avoid the need to explicitly articulate each predator-prey interactions as they and can be estimated from the commonly collected measures of abundance and size. The “size spectrum” describes the distribution of biomass or abundance as a function of individuals’ mass or size on a log–log scale (Guiet et al. (2016); Kerr and Dickie (2001)) . Size spectra are described by two terms, the size spectrum slope & intercept. These two terms reveal a sense of the baseline productivity, and how energy flows through an ecosystem (in the form of biomass) from many smaller individuals to many fewer large individuals. The spectrum intercept has been linked to the productivity of the community, and is often connected to the prevailing environmental conditions (Boudreau and Dickie (1992); Rossberg (2012)).

How are they used in practice:

Size spectra condense the complexities of predator prey networks and their interactions into a handful of size-based indices. These indices capture the emergent properties of a system, and have become increasingly used as indicators of ecosystem health. Within the context of fisheries management, changes in spectrum slopes have been associated with fishing exploitation, primarily through the targeted removal of larger individuals (Bianchi et al. (2000); Shin et al. (2005)). Numerical experiments have also linked changes in slope to environmental disturbances (Guiet et al. (2016)). Size spectra have also been shown to express predictable relationships between ecosystems of similar productivity levels as well as from distinct temperatures (Guiet et al. (2016)).

Use Pomerantz paper & Edwards to extend into ISD

Temperature & Ecology

Temperature plays a critical role on biological life impacting many of the chemical reactions that underpin basic physiological function. Temperature has direct and indirect impacts on critical biological functions including the acquisition of biomass through feeding, the rates of biomass loss through metabolism, and the rates at which individuals mature and develop. Because of these relationships, most species have evolved thermal preferences around which these chemical reactions are optimized. Species that are unable to maintain their thermal preferences internally must be able to follow their thermal preference in the environment through locomotion or adapt to less-favorable conditions through changes in behavior or risk metabolic costs in failing to do so. In an era of anthropogenic climate change, there is an expectation that many species will be displaced from historic habitats in their efforts to follow their thermal preferences. Recent research in marine environments has shown evidence of this as species are now shifting to higher latitudes and to deeper depths in the pursuit of more favorable conditions (Kleisner et al. (2017); Pinsky et al. (2013)). Others have suggested that temperature related impacts may not be seen through geographic distribution change, but through physiological changes, changes in seasonal phenology, or in dashed hopes of species recovery (Daan et al. (2005); Miller et al. (2018); Pershing et al. (2015); University of South Carolina et al. (2021)).

Need to connect temperature to size here

Direct quote from Guiet et al. (2016) , but nails the connection back to temp expectations:

Because it controls chemical reactions, temperature controls metabolic rates which underpin maintenance, growth or reproduction (Clarke and Johnston, 1999; Kooijman, 2010) as well as the functional responses to food density (Rall et al., 2012). Guiet et al. (2016)… In addition to the impact of temperature on communities’ intercepts (heights), the impact of temperature on the speed of the energy flow within communities may affect other properties, such as their resilience to perturbations or the intensity of trophic cascades (Andersen and Pedersen, 2009).

The potential for elevated temperatures to impact the size structure of an ecosystem has implications for the ecosystem resilience in the face of climate change, as well as the blue economies & natural resource systems that rely upon their good health.

Temperature of the Gulf of Maine & NE Shelf

In addition to the ecological disturbances of industrial fisheries, the Northwest Atlantic is also one of the fastest warming locations in the global oceans. Sea surface temperatures in the Gulf of Maine since 1982 have been warming at rates faster than 96% of the world’s oceans, with similar warming rates along the northwest Atlantic shelf (Pershing et al. (2018)). This persistent elevated temperature regime of the area is a result of several forces, a combination of shifting ocean currents and the unique bathymetry of the region. A Northward shift in the Gulf Stream directly increased the regional temperatures through increased transport of warm Gulf Stream water into areas like the Gulf of Maine. The Northward Gulf Stream shift is associated with a higher frequency of warm core rings, and the obstruction of cold-water Scotian Shelf current flow that would otherwise counter the influence of the Gulf Stream on the region’s temperatures (Gangopadhyay et al. (2019); University of South Carolina et al. (2021)). The combination of these oceanographic changes has led to a warmer continental shelf habitat.

The rapid warming in the northwest Atlantic is a major factor in the redistribution of marine species along the US east coast. Species have responded by adjusting the timing and locations of their seasonal migrations and shifting their geographic ranges (Nye et al. (2009); Staudinger et al. (2019)). There is evidence that warming has hampered fisheries recoveries as well. Adding a metabolic tax to physiological pathways like growth and metabolism. Species like Black Sea Bass, Atlantic shortfin squid, and Blue crab have been high-profile examples of species expanding their ranges to follow their thermal preferences. While species like the American lobster have shown declines at their southern range near Long Island Sound, with much doubt whether they will recover under the present temperature trends. The recent regime shift in the physical oceanography has also shown to be a catalyst for biological shifts as well (University of South Carolina et al. (2021); Perretti et al. (2017)).

While these examples show that species can respond to changes in the physical environment around them through movement & behavior, research elsewhere suggests that physiological responses integrated across species will manifest as changes in community size structure.

Fishing Impacts on Size/Species Composition

NEED fishing impacts on size structure references

Purpose

With the understanding that populations depend on the health of their ecosystems, there is a need to have community-wide metrics to effectively understand and manage marine resources. Size based indices are metrics that can be estimated from the information that has historically been available from long-term survey efforts. These indices have been shown to be sensitive to the impacts of fishing, but should also capture environmentally driven stress as well. We estimated size spectrum relationships as SBI’s for the groundfish populations for each sub-region of the Northeast US continental shelf. In the case of the NW Atlantic sustained increases in temperature should have a physiological impact on the community size structure.

This leads to our second hypothesis:

H2. Warming alters the community through the direct influence of temperature on metabolism, growth, and population productivity.

Methods

Fish Data Source and Processing

Data on the biomass, abundance, and size of fish on the Northeast U.S. Shelf were collected as part of the Northeast Fisheries Science Center’s bottom trawl survey ( (grosslein1969?), (Azarovits1981?), (politis2014?) ) This survey is conducted from Cape Hatteras, North Carolina to the Gulf of Maine each year in the spring and in the fall.  The survey follows a stratified random sampling design, with strata defined based on depth, bottom habitat, and latitude. Trawls are performed for a fixed duration at each station, reporting aggregate abundance and biomass for all species caught, and measuring individual lengths and weights for the catch of each species or a sub-sample if that catch is large. Correction factors were applied to aggregate species abundance and biomass to account for changes in vessels, gear, and doors used in the survey over time ( (sissenwine1978?), (byrne1991?), Miller et al. (2010) ) However, abundance and biomass at length needed to be estimated after these aggregate corrections. As such, abundance at length for each species was adjusted to match the corrected aggregate species abundance at each station, such that for each species, the sum of the resulting estimated abundance numbers across each length is equal to the corrected aggregate abundance.

Community Composition & Functional Groups

Analyses were performed using 68 species. These species were selected based on the availability of published weight-at-length relationships (Wigley et al. 2003) and represented 98.98% of the total biomass caught in the survey. Each species was assigned to a functional group based on life history and geography using the definitions of (Hare et al. (2016)). Functional groups included coastal fish, diadromous fish, elasmobranch, groundfish, and pelagic fish (Table 1). Six species with available length-weight details did not have a functional group designation, these species were designated as reef species. Exploratory analyses showed that the pelagic species biomass was low in all regions, and is unlikely to be representative of true biomass trends due to gear selectivity.

Published length-weight relationships ( Wigley et al. 2003 ) were used to convert from length data, available for all individuals, into their corresponding biomass-at-length. To account for differences in sampling effort among survey strata, all corrected abundance-at-length data were area-stratified. Area-stratified biomass-at-length values were then computed as the product of area-stratified abundance-at-length and estimated weight-at-length. All analyses were performed using area-stratified abundances and their associated area-stratified biomass estimates, hereby referred to as simply abundance-at-length & biomass-at-length

Community Composition Metrics

Our analyses used all data collected during the spring and fall surveys from 1970-2019. Data were grouped using survey-design strata into four sub-regions: Gulf of Maine, Georges Bank, Southern New England, Mid-Atlantic Bight. These sub-regions have been widely used in regional ecological studies (e.g., ). For each region, we developed several time series indicators:

  1. Community Composition Metrics (regional abundance and biomass by functional group)

  2. Average length and weight of the aggregate community and each functional group

  3. Exponent of the Individual Size Distribution Spectra

Body Size Changes

The average body length (cm) and body weight (kg) was calculated for all catch for each region and within each functional group. Each body-size metric was weighted by the abundance-at-length of a species sampled at each station such that.

\[X_j = \frac{ \sum_{}^{} n_iX_{ij} }{ \sum~w_i }\] Where \(X\) is the body-size metric, \(j\) is the year, & \(n\) is the abundance for each station \(i\).

Data for body size trends were not truncated using any minimum or maximum size and reflect all available catch data for the 68, for which biomass-at-length could be estimated.

Estimating Size Spectra

Lengths of individuals in the catch data are measured to the nearest cm, with smaller specimens measured to the nearest millimeter. Because individual biomass is estimated from those length measurements, there is a range of possible body mass values between the cm & mm increments. The relationship between length and mass in fishes is exponential and taxon specific, so biases resulting from using only the lower or upper end of those ranges is different for each taxon and increases for larger taxa. To account for this and reduce biases, we used the extended likelihood method (MLEbin) of Edwards et al. (2020). This method estimates the exponent of size spectra (b) for a bounded power law relationship between abundance and our length-derived estimates of individual biomass.

The exponent of this relationship is analogous to slope estimates from linear regression on logarithmic axes, with a steeper slope (i.e. a more negative b) indicating fewer large-bodied individuals and/or more small-bodied fishes Carvalho et al. (2021). This method has been shown to be the most accurate for estimating the exponent of size spectra when tested against alternative methods that require arbitrary decisions around binning the data (White et al. (2008); Sprules and Barth (2016); Edwards et al. (2017) ).

\[ \begin{align*} f(x) = \frac{ (\lambda + 1)x^{\lambda} }{ x^{\lambda+1}_{max} - x^{\lambda+1}_{min} }~~~~~~\lambda\neq1, \\ \\ f(x) = \frac{1}{logx_{max} - logx_{min}}~~~~~~\lambda=1 \end{align*} \tag{1}\]

Using this method, the size spectra exponent (b) was estimated for each survey region from 1970 to 2019. A minimum biomass of 1g was used for the lower limit and a maximum biomass of 10kg was used as an upper limit for the ISD’s bounded power law probability density function Equation 1, where \(x\) is body mass & \(\lambda\) is the scaling exponent of the ISD. The biomass within these limits constitutes 97.83% of all estimated biomass for the 68 used in this study. This truncation accounts for poor gear selectivity at the smallest and largest size ranges and imposes shared limits to the size distribution we’d expect to sample across these different areas. Exponents of size spectra were calculated using code modified from the sizeSpectra package in R (Edwards et al. (2017); Edwards et al. (2020)).

Drivers of Size Spectrum Change

The impact of external factors on the changes in size spectra was correlated against several hypothesized driving forces related to both environmental regimes and anthropogenic disturbances. Potential large-scale environmental drivers include regional SST anomalies & the larger-scale impact of the Gulf Stream Index (GSI). Fishing pressure represents the primary top-down anthropogenic driver in the region, and was investigated using the aggregate regional commercial landings data. The potential bottom-up impacts of zooplankton community differences were incorporated at a regional scale using small and large zooplankton indices from the EcoMon plankton survey.

Gulf Stream Index

Gulf stream index (GSI) values were obtained from the ecodata package in R (Bastille & Hardison 2018) . This package supplies GSI data at monthly intervals following the methodology of Pérez-Hernández and Joyce (2014) and Joyce et al. (2019) , using as sea level height anomaly data from the Copernicus Marine Environment Monitoring Service

Sea Surface Temperature Data

Global Sea surface temperature data was obtained via NOAA’s optimally interpolated SST analysis (OISSTv2), providing daily temperature values at a 0.25° latitude x 0.25° longitude resolution (Reynolds et al. 2007). A daily climatology for every 0.25° pixel in the global data set was created using average daily temperatures spanning the period of 1982-2011. Daily anomalies were then computed as the difference between observed temperatures and the daily climatological average. OISSTv2 data used in these analyses were provided by the NOAA PSL, Boulder, Colorado, USA from their website at https://psl.noaa.gov/data/gridded/data.noaa.oisst.v2.html.

Sea surface temperature data was regionally averaged to match the survey regions from the age-at-length data. SST anomalies were averaged by year for each region and over the entire sampling region to produce daily time series. These time series were then processed into annual timeseries of surface temperatures and anomalies. All region-averaging was done with area-weighting of the latitude/longitude grid cells to account for differences in cell-size in the OISSTv2 data.

Commercial Fishing

Fishing pressure in the region was indexed using state and federal commercial fishing landings.  These data were obtained from the Greater Atlantic Regional Fisheries Office (GARFO) for statistical areas that are routinely used for fisheries reporting and management (Figure X).  Individual statistical areas were aggregated into regions that closely align with the survey areas we defined for the size spectra analyses (Figure X).

Zooplankton Abundance Indices

Zooplankton data are from the National Oceanographic and Atmospheric Administration Marine Resources Monitoring, Assessment and Prediction (MARMAP) program and Ecosystem Monitoring (EcoMon) cruises detailed extensively in Kane (2007), Kane (2011), and Morse et al. (2017). Abundance anomalies are computed from the expected abundance on the day of sample collection. The small copepod index is computed by averaging the individual abundance anomalies of Pseudocalanus spp., Centropages hamatus, Centropages typicus, and Temora longicornis. The large-copepod anomalies is the abundance anomaly of Calanus finmarchicus, the largest copepod in the Northeast U.S. region. Anomalies for both large and small zooplankton groups are then averaged within ecological production units for annual timeseries.

Analysis of Driver Impacts

Driver Cross Correlation Functions

In addition to evaluating the direct and contemporaneous impacts of each potential driver, additional predictors were identified for inclusion in the model-selection process through the use of cross correlation function (CCF) estimates and changepoint analysis. CCF estimates were used to look at correlation at one-year lag intervals to investigate potential time-delayed effects between the hypothesized drivers and the community’s size spectra. CCF estimates were performed between the dependent variables of the size spectra slope with the independent variables of the annual gulf stream index, the corresponding regional commercial landings and sea surface temperature anomalies, and both small and large zooplankton indices. Annual time lags on independent variables of up to 10 years that were identified in the CCF analysis were then included as additional candidate predictors in the regression analysis on drivers of the spectrum slope changes.

Slope Changepoint Analyses

Changepoint analysis was used as a means to explore the temporal structure in changes of the regional size spectrum slopes. Changepoint analysis was performed using the envcpt package ((killick?) et al), testing comparing support for 12 candidate model structures and any identified changepoints. The most supported changepoint structures were then used to inform whether (if any) regional spectra expressed distinct periods of behavior. If changepoints were identified, this was used as evidence that there may have been a community regime change. Following this thinking it is plausible for there to be inconsistencies or non-stationarity in the predictors’ impact on the community spectra, as different community regimes may be differently vulnerable to top-down pressures like commercial fishing. To allow for changes in these relationships across different regimes, separated by changepoints, an additional interaction term was added to the regression model candidates for that region.

Driver Correlation Periods

Regression Analyses

Following exploratory analyses with CCF functions and changepoint analyses, regression models were developed and tested independently for each region. Potential drivers included in the models were the annual Gulf Stream Index, the annual regionally averaged SST anomaly value, regionally averaged small and large copepod indices, and the scaled annual commercial landings from that region. Gulf stream index, zooplankton indices, and SST anomalies were not standardized prior to use in model fitting and evaluation as the two former are already an index value and the latter is scaled to a thirty-year climatological reference period. In the event that CCF analysis identified an important time lag for a region’s independent variables, these specific time-lagged predictors were included as predictors in the candidate models, otherwise time-lags on predictors were not evaluated. For any region(s) that showed support for changepoints in their temporal structure, the periods separated by those breaks were included as an interaction term on all other predictors. In the absence of any identified changepoints, no multi-year periods were included as potential regime impacts. All candidate models for a region were evaluated using AIC to determine the most parsimonious models for each regional system.

Results

Community Abundance

Stratified abundance estimates have been gradually increasing on the Northeast Shelf since 1970. Abundance growth accelerated after 2007 to peak levels in 2014. Abundance then began declining, which continued through 2019 (Figure 2). In the Gulf of Maine, fish abundance remained relatively-low until the 1990’s, at which point it began to steadily rise–closely aligning with the pattern for the Northeast Shelf. This increase in abundance reversed briefly during 2002-2006, but continued to rise after 2006 before hitting a regional peak in 2016. Populations then declined through 2019 similar to the shelf-wide trend. Georges Bank abundances were consistently low from 1970 around 2010. By 2014 abundance had roughly quadrupled, propelled by strong recruitment classes of haddock. After 2014, abundance quickly fell to numbers more similar to the 1970-2010 levels by the end of the decade. Community abundance in Southern New England displayed higher inter-annual changes across all years compared to both Gulf of Maine and Georges Bank. Abundance in this area showed a less dramatic rise and fall than the Northern regions. Abundances began increasing rapidly here in 2007, before falling back to earlier levels by the end of the 2010’s decade. The Mid Atlantic Bight displayed the most inter-annual variability and had no major trends in fish abundance.

Groundfish species were the dominant functional group driving the abundance and biomass trends in the Gulf of Maine and Georges Bank, with the two southern regions showing a more balanced abundance distribution among the five functional groups (Figure 2).

Community Biomass

Similar to abundance, the overall biomass was highest in the two northern regions, the Gulf of Maine and Georges Bank. Roughly half of the biomass sampled in these regions can be attributed to groundfish species, with the second largest contributions coming from elasmobranchs. Within the groundfish biomass, larger individuals >2kg in particular, declined during the 70’s and 80’s in these regions, never truly recovering. Beginning in the 2000’s there were signs that groundfish abundances were increasing as evidenced by increasing numbers of smaller individuals, however in both regions this trend appears to have reversed by the mid 2010’s. Elasmobranch biomass increased steadily throughout the survey time period across all regions, with the exception of southern New England. This area showed large 5-10 year swings in biomass, but no clear long-term trend. Larger elasmobranch were rare in all regions except for a period spanning the late 70’s through the early 90’s isolated to Georges Bank. Demersal species biomass was highest in the Gulf of Maine, dwarfing their contributions in other regions. Their biomass declined in the 70’s, was flat until the late 90’s, remaining relatively high until declining in the late 2010’s. Pelagic species biomass was low in all regions, and is unlikely to be representative of true biomass trends due to gear selectivity.

Regional Variation in Species Composition

There was a distinct difference between Northern and Southern regions in the way biomass was distributed among the different functional groups. The largest contributors to biomass in the southern regions (southern New England & mid-Atlantic bight) was the elasmobranch community. While the northern regions (Gulf of Maine & Georges Bank) each had similar quantities of elasmobranch biomasses, there was also a comparable contribution of groundfish and in the Gulf of Maine there was a major component of demersal species as well.

Body Size Trends

The average fish size in the Gulf of Maine (length and weight) declined the greatest of all regions over our study period. The average individual length was greatest in the 1970’s in the 35-40cm range, falling to 28-33cm over the last decade. Body-weight fell dramatically in the 1980’s, from around .75kg in the 1970’s to .25-.30kg, roughly a third of what it had been. Georges Bank body sizes also declined during the study period, but less dramatically. Both of these Northern regions had brief period in the late 2000’s where average length and weight rose, before falling again in the 2010’s. The MAB region was the only region to see a long-term increase in both length and weight during the study period. SNE saw no long-term change in length, and a minor decline in average body-weight.

Regional Size Spectra

At the start of our time series, back in the 1970’s, there was a clear difference in the relative positions of spectra parameters among the different regions. Gulf of Maine and Georges Bank showed the least steep spectra slopes in the earlier time periods with slopes around -1 & -1.1 respectively. The relatively flat slopes in these regions both steepened over time, settling near -1.3 (GoM) and -1.5 (GB). Gulf of Maine experienced much of its decline during the 1980’s and 1990’s. There was a brief reversal in this trend during the 2000’s, but slopes continued to steepen by 2010 and remained steep through 2019. Georges Bank did not experience as rapid of a decline, but experienced a similar long-term steepening. In contrast to the northern regions, SNE and MAB had steeper slopes in the -1.2 to -1.5 territory. The long term pattern for SNE was one of increasing volatility, but not so much a decline. The spectra slope for the MAB was less volatile, but similarly maintained a relatively stable wander around -1.4. By the end of the study period all regions had slopes that were at or near a similar level.

Size Spectra Drivers

Spectra Slope changepoints and Regression Analysis

Following break point exploratory analyses with the EnvCpt package, the Gulf of Maine was the only region from which changepoints were identified in the best supported model from the candidate changepoint model structures. Changes in Gulf of Maine’s spectrum slope was best modeled by a model allowing breaks in the trend, and with an auto-regressive process on the previous 2 years. From this model, breaks were identified at 1999 and 2007. Georges Bank was best modeled with a single trend, with no changepoints. Southern New-England showed the most support for a single intercept model, with no changepoints, with values centered on mu = -1.04 & sd = 0.046. The Mid-Atlantic Bight was best supported by a single trend model, with no changepoints, and an increasing intercept through the study period.

Region Most Supported Model
gom trendar2cpt
gb trend
sne meancpt
mab trend

Driver Differences Across Regimes

changepoint analyses suggest there were as many as 3 distinct regimes in the Gulf of Maine slope changes. For all other regions there was not strong support for changepoints.

Spectra Slope Regression Analysis

Characteristic Beta 95% CI1 p-value
blag2 -0.33 -0.64, -0.03 0.033
landings 0.06 0.01, 0.10 0.009
regime
    yrs_1982-1998 — —
    yrs_1999_2006 -0.04 -0.15, 0.06 0.4
    yrs_2007_2019 -1.9 -4.1, 0.40 0.10
zp_large -0.06 -0.17, 0.04 0.2
zp_small -0.07 -0.12, -0.02 0.011
landings * regime
    landings * yrs_1999_2006 -0.20 -0.38, -0.02 0.032
    landings * yrs_2007_2019 -2.0 -4.4, 0.33 0.088
regime * zp_large
    yrs_1999_2006 * zp_large 0.01 -0.18, 0.19 >0.9
    yrs_2007_2019 * zp_large 0.40 0.18, 0.62 <0.001
R² 0.653
Adjusted R² 0.515
Sigma 0.034
Statistic 4.71
p-value <0.001
df 10
Log-likelihood 77.3
AIC -131
BIC -112
Deviance 0.029
Residual df 25
No. Obs. 36
1 CI = Confidence Interval
Predictor GVIF Df GVIF^(1/(2*Df)) Interacts With Other Predictors
regime 1.000000e+00 17 1.000000 SST, GSI, landings, zp_small, zp_large --
SST 5.960859e+07 5 5.991435 regime GSI, landings, zp_small, zp_large
GSI 2.074176e+07 5 5.391189 regime SST, landings, zp_small, zp_large
landings 4.268892e+03 5 2.306915 regime SST, GSI, zp_small, zp_large
zp_small 1.460115e+08 5 6.552978 regime SST, GSI, landings, zp_large
zp_large 1.918367e+07 5 5.349253 regime SST, GSI, landings, zp_small

Exploratory analyses for Georges Bank indicated that the structure of the data did not show support for clear changepoints. CCF results of individual predictors showed that spectrum slope was correlated with small zooplankton index values from two years prior. This relationship was added to the standard predictor candidates, each operating at a simultaneous time step.

Characteristic Beta 95% CI1 p-value
SST -0.03 -0.06, 0.00 0.028
zpslag2 -0.08 -0.13, -0.02 0.006
R² 0.307
Adjusted R² 0.264
Sigma 0.052
Statistic 7.09
p-value 0.003
df 2
Log-likelihood 55.6
AIC -103
BIC -97.0
Deviance 0.085
Residual df 32
No. Obs. 35
1 CI = Confidence Interval
Predictor VIF
landings 1.929115
zpslag1 2.791261
SST 3.077146
zp_large 3.265561
zp_small 3.887965
zpslag2 4.634856
GSI 4.918679

Characteristic Beta 95% CI1 p-value
landings 0.01 0.00, 0.03 0.13
R² 0.062
Adjusted R² 0.036
Sigma 0.046
Statistic 2.37
p-value 0.13
df 1
Log-likelihood 64.4
AIC -123
BIC -118
Deviance 0.075
Residual df 36
No. Obs. 38
1 CI = Confidence Interval
Predictor VIF
zp_large 1.507136
zp_small 1.621108
GSI 1.932934
landings 2.242318
SST 2.752092

Characteristic Beta 95% CI1 p-value
zp_small -0.07 -0.13, -0.01 0.020
R² 0.142
Adjusted R² 0.118
Sigma 0.060
Statistic 5.96
p-value 0.020
df 1
Log-likelihood 53.9
AIC -102
BIC -96.9
Deviance 0.130
Residual df 36
No. Obs. 38
1 CI = Confidence Interval
Predictor VIF
landings 1.331770
zp_large 1.373941
zp_small 1.636199
SST 1.763223
GSI 1.831599

Potential Drivers Timeseries:

The following panels show the historical changes in each of the drivers. Landings have been scaled by average total landings within each region across all years. SST and GSI have not been scaled. This is different from how they are implemented in the regression analyses, when the landings were scaled over the 1982-2019 period.

Supplemental Materials

Table 1: Common and scientific names for the species that constitute each functional group used in our analyses. X markers are used to indicate which regions each species has been caught in the data.
Functional Group Assignments and Regional Presence/Absence
Common Name Scientific Name Gulf of Maine Georges Bank Southern New England Mid-Atlantic Bight
Coastal - (18)
Atlantic Croaker micropogonias undulatus X X X
Atlantic Spadefish chaetodipterus faber X
Atlantic Thread Herring opisthonema oglinum X X
Black Sea Bass centropristis striata X X X X
Blackbelly Rosefish helicolenus dactylopterus X X X X
Blueback Herring alosa aestivalis X X X X
Bluefish pomatomus saltatrix X X X X
Butterfish peprilus triacanthus X X X X
Cunner tautogolabrus adspersus X X X X
Greater Amberjack seriola dumerili X X
Northern Kingfish menticirrhus saxatilis X X X
Scup stenotomus caprinus X X X X
Southern Kingfish menticirrhus americanus X
Spanish Mackerel scomberomorus maculatus X
Spanish Sardine sardinella aurita X
Spot leiostomus xanthurus X X
Striped Bass morone saxatilis X X X X
Weakfish cynoscion regalis X X X
Diadromous - (2)
American Shad alosa sapidissima X X X X
Atlantic Sturgeon acipenser oxyrhynchus X
Elasmobranch - (19)
Atlantic Angel Shark squatina dumeril X
Atlantic Sharpnose Shark rhizoprionodon terraenovae X
Barndoor Skate dipturus laevis X X X X
Bullnose Ray myliobatis freminvillei X X
Chain Dogfish scyliorhinus retifer X X X
Clearnose Skate raja eglanteria X X
Cownose Ray rhinoptera bonasus X
Little Skate leucoraja erinacea X X X X
Rosette Skate leucoraja garmani X X X X
Roughtail Stingray dasyatis centroura X
Sand Tiger carcharias taurus X
Sandbar Shark carcharhinus plumbeus X X
Smooth Butterfly Ray gymnura micrura X
Smooth Dogfish mustelus canis X X X X
Smooth Skate malacoraja senta X X X X
Spiny Butterfly Ray gymnura altavela X
Spiny Dogfish squalus acanthias X X X X
Thorny Skate amblyraja radiata X X X X
Winter Skate leucoraja ocellata X X X X
Groundfish - (25)
Acadian Redfish sebastes fasciatus X X X X
American Plaice hippoglossoides platessoides X X X X
Atlantic Cod gadus morhua X X X X
Atlantic Halibut hippoglossus hippoglossus X X X
Atlantic Wolffish anarhichas lupus X X X
Cusk brosme brosme X X X X
Fawn Cusk-Eel lepophidium profundorum X X X X
Fourspot Flounder paralichthys oblongus X X X X
Goosefish lophius americanus X X X X
Haddock melanogrammus aeglefinus X X X X
Longhorn Sculpin myoxocephalus octodecemspinosus X X X X
Northern Searobin prionotus carolinus X X X X
Ocean Pout macrozoarces americanus X X X X
Offshore Hake merluccius albidus X X X X
Pollock pollachius virens X X X X
Red Hake urophycis chuss X X X X
Sea Raven hemitripterus americanus X X X X
Silver Hake merluccius bilinearis X X X X
Spotted Hake urophycis regia X X X X
Summer Flounder paralichthys dentatus X X X X
White Hake urophycis tenuis X X X X
Windowpane Flounder scophthalmus aquosus X X X X
Winter Flounder pseudopleuronectes americanus X X X X
Witch Flounder glyptocephalus cynoglossus X X X X
Yellowtail Flounder limanda ferruginea X X X X
Pelagic - (4)
Atlantic Herring clupea harengus X X X X
Atlantic Mackerel scomber scombrus X X X X
Buckler Dory zenopsis conchifera X X X X
Round Herring etrumeus teres X X X X
Functional group assignments adapted from Hare et al. 2010
Top Commercial Fisheries Landings of Northeastern US (by weight)
Avg. Annual Landings (lb.) Total Landings (lb.) Total Value ($)
Gulf of Maine - 1960
Hake, Silver 16.58M 281.87M 8.71M
Herring, Atlantic 11.57M 138.83M 2.50M
Redfish, Acadian 2.12M 88.97M 3.41M
Gulf of Maine - 1970
Herring, Atlantic 22.78M 501.08M 19.70M
Menhaden, Atlantic 17.78M 373.48M 7.87M
Redfish, Acadian 3.14M 219.85M 23.87M
Gulf of Maine - 1980
Herring, Atlantic 21.78M 653.26M 34.52M
Menhaden, Atlantic 21.24M 509.75M 12.77M
Pollock 3.33M 229.57M 62.00M
Gulf of Maine - 1990
Herring, Atlantic 25.21M 958.12M 54.12M
Cod, Atlantic 2.35M 138.76M 131.76M
Shark, Dogfish, Spiny 3.34M 120.17M 15.95M
Gulf of Maine - 2000
Herring, Atlantic 2.99M 47.77M 4.31M
Monkfish/Angler/Goosefish 716.21K 31.51M 51.13M
Cod, Atlantic 692.95K 30.49M 42.30M
Gulf of Maine - 2010
Tuna, Bluefin 209.06K 3.76M 33.30M
Shark, Dogfish, Spiny 479.11K 2.87M 590.62K
Pollock 188.20K 1.69M 2.08M
Georges Bank - 1960
Haddock 15.00M 270.06M 34.41M
Hake, Silver 6.83M 95.57M 3.19M
Cod, Atlantic 4.88M 87.89M 8.12M
Georges Bank - 1970
Cod, Atlantic 7.78M 233.48M 59.16M
Flounder, Yellowtail 4.62M 138.52M 43.16M
Redfish, Acadian 2.63M 76.37M 9.09M
Georges Bank - 1980
Cod, Atlantic 10.11M 404.40M 211.60M
Flounder, Winter 2.50M 100.11M 89.84M
Haddock 2.36M 94.27M 66.68M
Georges Bank - 1990
Cod, Atlantic 4.27M 192.29M 190.26M
Hake, Silver 1.79M 76.82M 20.49M
Flounder, Winter 1.23M 56.43M 75.59M
Georges Bank - 2000
Cod, Atlantic 2.17M 62.91M 75.20M
Herring, Atlantic 3.49M 48.92M 3.73M
Haddock 1.54M 43.01M 55.37M
Georges Bank - 2010
Hake, Silver 155.88K 779.40K 499.90K
Haddock 39.65K 118.95K 143.04K
Flounder, Winter 40.40K 80.80K 216.28K
Southern New England - 1960
Other Fish, Bony 14.84M 400.77M 3.73M
Flounder, Yellowtail 6.56M 196.83M 19.12M
Flounder, Winter 2.52M 70.58M 7.01M
Southern New England - 1970
Menhaden, Atlantic 9.99M 239.84M 5.12M
Other Fish, Bony 4.05M 206.59M 2.49M
Flounder, Yellowtail 2.07M 153.55M 36.47M
Southern New England - 1980
Menhaden, Atlantic 6.60M 217.68M 10.21M
Hake, Silver 2.56M 205.02M 46.11M
Flounder, Yellowtail 1.66M 132.92M 83.38M
Southern New England - 1990
Hake, Silver 2.52M 196.81M 78.54M
Herring, Atlantic 2.12M 129.02M 7.19M
Menhaden, Atlantic 3.71M 125.98M 8.69M
Southern New England - 2000
Mackerel, Atlantic 2.55M 135.06M 15.60M
Hake, Silver 1.00M 55.25M 26.89M
Skate, Nk 950.56K 49.43M 6.53M
Southern New England - 2010
Scup 161.10K 6.44M 4.29M
Hake, Silver 145.07K 4.21M 3.12M
Flounder, Summer 80.43K 3.86M 11.54M
Mid-Atlantic Bight - 1960
Flounder, Summer 2.03K 4.05K 720.00
Flounder, Yellowtail 2.33K 2.33K 214.00
Flounder, Witch 395.00 395.00 36.00
Mid-Atlantic Bight - 1970
Menhaden, Atlantic 10.20M 50.98M 1.59M
Weakfish/Sea Trout, Squeteague 886.91K 9.76M 1.40M
Scup 876.60K 8.77M 2.09M
Mid-Atlantic Bight - 1980
Menhaden, Atlantic 30.78M 646.41M 10.94M
Flounder, Summer 1.15M 83.83M 72.00M
Scup 550.89K 37.46M 15.53M
Mid-Atlantic Bight - 1990
Menhaden, Atlantic 115.86M 4.63B 286.14M
Mackerel, Atlantic 1.67M 103.62M 13.87M
Croaker, Atlantic 1.35M 71.65M 22.53M
Mid-Atlantic Bight - 2000
Menhaden, Atlantic 69.60M 2.64B 167.17M
Croaker, Atlantic 2.16M 106.02M 42.93M
Mackerel, Atlantic 1.70M 59.41M 6.38M
Mid-Atlantic Bight - 2010
Menhaden, Atlantic 118.29M 1.89B 154.46M
Bass, Striped 1.70M 25.56M 75.05M
Croaker, Atlantic 1.08M 24.81M 21.37M
Landings data obtained from the Greater Atlantic Regional Fishing Office (GARFO)

References

Bertalanffy, L. von. 1938. A quantitative theory of organic growth (inquiries on growth laws. II.). Human Biology, 10: 181213. https://www.proquest.com/docview/1301827069/citation/94DE10A3AF8F4DF7PQ/1.
Bertalanffy, L. von. 1957. Quantitative laws in metabolism and growth. The Quarterly Review of Biology, 32: 217–231. https://www.journals.uchicago.edu/doi/abs/10.1086/401873.
Bianchi, G., Gislason, H., Graham, K., Hill, L., Jin, X., Koranteng, K., Manickchand-Heileman, S., et al. 2000. Impact of fishing on size composition and diversity of demersal fish communities. ICES Journal of Marine Science, 57: 558–571. https://doi.org/10.1006/jmsc.2000.0727.
Boudreau, P. R., and Dickie, L. M. 1992. Biomass spectra of aquatic ecosystems in relation to fisheries yield. Canadian Journal of Fisheries and Aquatic Sciences, 49: 1528–1538. https://cdnsciencepub.com/doi/abs/10.1139/f92-169.
Brown, J., West, G. B., and Enquist, B. 2000. Scaling in biology: Patterns and processes, causes and consequences. In pp. 1–24.
Carvalho, P. G., Setiawan, F., Fahlevy, K., Subhan, B., Madduppa, H., Zhu, G., and Humphries, A. T. 2021. Fishing and habitat condition differentially affect size spectra slopes of coral reef fishes. Ecological Applications, 31: e02345. https://onlinelibrary.wiley.com/doi/abs/10.1002/eap.2345.
Currie, D. J. 1993. What shape is the relationship between body size and population density? Oikos, 66: 353. https://www.jstor.org/stable/3544825?origin=crossref.
Daan, N., Gislason, H., G. Pope, J., and C. Rice, J. 2005. Changes in the North Sea fish community: evidence of indirect effects of fishing? ICES Journal of Marine Science, 62: 177–188. https://academic.oup.com/icesjms/article/62/2/177/602590.
Damuth, J. 1981. Population density and body size in mammals. Nature, 290: 699–700. http://www.nature.com/articles/290699a0.
Edwards, A., Robinson, J., Blanchard, J., Baum, J., and Plank, M. 2020. Accounting for the bin structure of data removes bias when fitting size spectra. Marine Ecology Progress Series, 636: 19–33. https://www.int-res.com/abstracts/meps/v636/p19-33/.
Edwards, A. M., Robinson, J. P. W., Plank, M. J., Baum, J. K., and Blanchard, J. L. 2017. Testing and recommending methods for fitting size spectra to data. Methods in Ecology and Evolution, 8: 57–67. https://besjournals.onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.12641.
Fogarty, M. J., and Murawski, S. A. 1998. Large-Scale Disturbance and the Structure of Marine Systems: Fishery Impacts on Georges Bank. Ecological Applications, 8: S6–S22. https://onlinelibrary.wiley.com/doi/abs/10.1890/1051-0761%281998%298%5BS6%3ALDATSO%5D2.0.CO%3B2.
Gangopadhyay, A., Gawarkiewicz, G., Silva, E. N. S., Monim, M., and Clark, J. 2019. An Observed Regime Shift in the Formation of Warm Core Rings from the Gulf Stream. Scientific Reports, 9: 12319. http://www.nature.com/articles/s41598-019-48661-9.
Gillooly, J. F., Brown, J. H., West, G. B., Savage, V. M., and Charnov, E. L. 2001. Effects of Size and Temperature on Metabolic Rate. Science, 293: 2248–2251. https://www.science.org/doi/10.1126/science.1061967.
Guiet, J., Poggiale, J.-C., and Maury, O. 2016. Modelling the community size-spectrum: recent developments and new directions. Ecological Modelling, 337: 4–14. https://www.sciencedirect.com/science/article/pii/S0304380016301958.
Hare, J. A., Morrison, W. E., Nelson, M. W., Stachura, M. M., Teeters, E. J., Griffis, R. B., Alexander, M. A., et al. 2016. A Vulnerability Assessment of Fish and Invertebrates to Climate Change on the Northeast U.S. Continental Shelf. PLOS ONE, 11: e0146756. https://dx.plos.org/10.1371/journal.pone.0146756.
Heatwole, H., Torres, F., Blasini De Austin, S., and Heatwole, A. 1969. Studies on anuran water balanceI. Dynamics of evaporative water loss by the coquí, eleutherodactylus portoricensis. Comparative Biochemistry and Physiology, 28: 245–269. https://www.sciencedirect.com/science/article/pii/0010406X69913425.
Hillaert, J., Hovestadt, T., Vandegehuchte, M. L., and Bonte, D. 2018. Size-dependent movement explains why bigger is better in fragmented landscapes. Ecology and Evolution, 8: 10754–10767. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6262741/.
Joyce, T. M., Kwon, Y.-O., Seo, H., and Ummenhofer, C. C. 2019. Meridional Gulf Stream Shifts Can Influence Wintertime Variability in the North Atlantic Storm Track and Greenland Blocking. Geophysical Research Letters, 46: 1702–1708. https://onlinelibrary.wiley.com/doi/abs/10.1029/2018GL081087.
Kerr, S. R., and Dickie, L. M. 2001. The Biomass Spectrum: A Predator-Prey Theory of Aquatic Production. Columbia University Press.
Kleisner, K. M., Fogarty, M. J., McGee, S., Hare, J. A., Moret, S., Perretti, C. T., and Saba, V. S. 2017. Marine species distribution shifts on the U.S. Northeast Continental Shelf under continued ocean warming. Progress in Oceanography, 153: 24–36. https://linkinghub.elsevier.com/retrieve/pii/S0079661116301896.
Loeuille, N., and Loreau, M. 2006. Evolution of body size in food webs: does the energetic equivalence rule hold? Ecology Letters, 9: 171–178.
Miller, K., Charles, A., Barange, M., Brander, K., Gallucci, V. F., Gasalla, M. A., Khan, A., et al. 2010. Climate change, uncertainty, and resilient fisheries: Institutional responses through integrative science. Progress in Oceanography, 87: 338–346. https://orbit.dtu.dk/en/publications/climate-change-uncertainty-and-resilient-fisheries-institutional-.
Miller, T. J., O’Brien, L., and Fratantoni, P. S. 2018. Temporal and environmental variation in growth and maturity and effects on management reference points of Georges Bank Atlantic cod. Canadian Journal of Fisheries and Aquatic Sciences, 75: 2159–2171. http://www.nrcresearchpress.com/doi/10.1139/cjfas-2017-0124.
Nye, J., Link, J., Hare, J., and Overholtz, W. 2009. Changing spatial distribution of fish stocks in relation to climate and population size on the Northeast United States continental shelf. Marine Ecology Progress Series, 393: 111–129. http://www.int-res.com/abstracts/meps/v393/p111-129/.
Pérez-Hernández, M. D., and Joyce, T. M. 2014. Two Modes of Gulf Stream Variability Revealed in the Last Two Decades of Satellite Altimeter Data. Journal of Physical Oceanography, 44: 149–163. https://journals.ametsoc.org/view/journals/phoc/44/1/jpo-d-13-0136.1.xml.
Perretti, C., Fogarty, M., Friedland, K., Hare, J., Lucey, S., McBride, R., Miller, T., et al. 2017. Regime shifts in fish recruitment on the Northeast US Continental Shelf. Marine Ecology Progress Series, 574: 1–11. http://www.int-res.com/abstracts/meps/v574/p1-11/.
Pershing, A., Mills, K., Dayton, A., Franklin, B., and Kennedy, B. 2018. Evidence for Adaptation from the 2016 Marine Heatwave in the Northwest Atlantic Ocean. Oceanography, 31. https://tos.org/oceanography/article/evidence-for-adaptation-from-the-2016-marine-heatwave-in-the-northwest-atl.
Pershing, A. J., Alexander, M. A., Hernandez, C. M., Kerr, L. A., Le Bris, A., Mills, K. E., Nye, J. A., et al. 2015. Slow adaptation in the face of rapid warming leads to collapse of the gulf of maine cod fishery. Science, 350: 809–812. https://www.science.org/doi/full/10.1126/science.aac9819.
Pinsky, M. L., Worm, B., Fogarty, M. J., Sarmiento, J. L., and Levin, S. A. 2013. Marine Taxa Track Local Climate Velocities. Science, 341: 1239–1242. https://science.sciencemag.org/content/341/6151/1239.
Pomeranz, J. P. F., Junker, J. R., and Wesner, J. S. 2022. Individual size distributions across North American streams vary with local temperature. Global Change Biology, 28: 848–858. https://onlinelibrary.wiley.com/doi/10.1111/gcb.15862.
Rossberg, A. G. 2012. 6 - A Complete Analytic Theory for Structure and Dynamics of Populations and Communities Spanning Wide Ranges in Body Size. In pp. 427–521. Ed. by U. Jacob and G. Woodward. Academic Press. https://www.sciencedirect.com/science/article/pii/B9780123969927000083.
Sheldon, R. W., Prakash, A., and Sutcliffe Jr., W. H. 1972. The Size Distribution of Particles in the Ocean1. Limnology and Oceanography, 17: 327–340. https://onlinelibrary.wiley.com/doi/abs/10.4319/lo.1972.17.3.0327.
Shin, Y.-J., Rochet, M.-J., Jennings, S., Field, J. G., and Gislason, H. 2005. Using size-based indicators to evaluate the ecosystem effects of fishing. ICES Journal of Marine Science, 62: 384–396. https://doi.org/10.1016/j.icesjms.2005.01.004.
Sprules, W. G., and Barth, L. E. 2016. Surfing the biomass size spectrum: some remarks on history, theory, and application. Canadian Journal of Fisheries and Aquatic Sciences, 73: 477–495. http://www.nrcresearchpress.com/doi/10.1139/cjfas-2015-0115.
Staudinger, M. D., Mills, K. E., Stamieszkin, K., Record, N. R., Hudak, C. A., Allyn, A., Diamond, A., et al. 2019. It’s about time: A synthesis of changing phenology in the Gulf of Maine ecosystem. Fisheries Oceanography, 28: 532–566. https://onlinelibrary.wiley.com/doi/abs/10.1111/fog.12429.
University of South Carolina, Meyer-Gutbrod, E., Greene, C., Davies, K., and Johns, D. 2021. Ocean Regime Shift is Driving Collapse of the North Atlantic Right Whale Population. Oceanography, 34: 22–31. https://tos.org/oceanography/article/ocean-regime-shift-is-driving-collapse-of-the-north-atlantic-right-whale-population.
White, E. P., Ernest, S. K. M., Kerkhoff, A. J., and Enquist, B. J. 2007. Relationships between body size and abundance in ecology. Trends in Ecology & Evolution, 22: 323–330. https://linkinghub.elsevier.com/retrieve/pii/S0169534707000985.
White, E. P., Enquist, B. J., and Green, J. L. 2008. On Estimating the Exponent of Power-Law Frequency Distributions. Ecology, 89: 905–912. https://onlinelibrary.wiley.com/doi/abs/10.1890/07-1288.1.